The prediction that keeps on being true
Back in 1965 Gordon Moore, a co-founder of Intel, predicted that the number of transistors per square inch on integrated circuits would double every year. Intel recently celebrated the 40 th anniversary of this prediction, which has become known as "Moores Law" because of its scary accuracy.
In computers, transistors are basically electrical switches that have two settings, on and off hence the ones and zeros of basic binary, the base language upon which all computing is built. These transistors are built into logic circuits, circuits are interconnected on wafers, the wafers are assembled into stacks, and stacks into processors.
Give or take, Moores predictions were almost perfect. Sometimes leaps in technology have allowed processor makers to more than double performance, while in recent years this doubling has taken 18 to 24 months as chipmakers are reaching the physical limitations of materials used to manufacture and assemble chips.
Were down to about 60 nanometers in width for circuits, down from 90 nanometers just a few years ago. A nanometer is one billionth of a meter, or one millionth of a millimeter, so dont even try to picture how thin that is.
Intel is hoping to produce a 45 nanometer chip for 2007, a 32 nanometer chip for 2009, and a 22 nanometer chip in 2011. Each thinning of circuitry will have corresponding benefits in processing power and miniaturization.
But while it might seem like Moores Law will inevitably slow and stop as we reach the physical limitations of various materials you can only stretch copper so thin new scientific research shows that, if anything, were about to shift Moores Law into overdrive.
Various techniques, such as nanotechnology and microdots are being investigated, and one Israeli company is developing fibre optic processors that use light-activated switches rather than electron-triggered transistors.
Recently two groups, one a Simon Fraser University lab, another a University of Arizona lab, performed two different experiments that could herald the creation of transistors based on a ring of simple molecules.
In a conventional microchip, it requires thousands and sometimes millions of electrons to turn a transistor on or off. A molecular transistor would require a single electron.
While it will be at least a decade before molecular computing could even be possible in any kind of commercial setting, this breakthrough could increase the number of transistors on a chip exponentially, improving the speed and performance of processors while significantly reducing chip size and chip heat.
While its difficult to predict what this will actually mean in terms of computing, it does reinforce an important corollary to Moores Law that no matter what or when you buy it, your system is doomed to become dated in a matter of months and obsolete in a matter of years.